Ousset, Aymeric, Bassand, Celine, Chavez, Pierre­Francois, Meeus, Joke, Robin, Florent, Schubert, Martin Alexander, Somville, Pascal and Dodou, Kalliopi (2018) Development   of   a   small­scale   spray­drying   approach   for   amorphous   solid dispersions   (ASDs)   screening   in   early   drug   development.   Pharmaceutical Development and Technology. 

Downloaded from: http://sure.sunderland.ac.uk/id/eprint/10154/

Usage guidelines

Please   refer   to   the  usage guidelines  at  http://sure.sunderland.ac.uk/policies.html  or  alternatively 

contact sure@sunderland.ac.uk.

Development of a small-scale spray-drying approach for amorphous

solid dispersions (ASDs) screening in early drug development

Aymeric Ousseta, Céline Bassandb, Pierre-François Chavezb, Joke Meeusb,

Florent Robinb, Martin Alexander Schubertb, Pascal Somvilleb, Kalliopi

Dodoua*

aSchool of Pharmacy and Pharmaceutical Sciences, Faculty of Health Sciences and

Wellbeing, University of Sunderland, Sunderland SR13SD, UK; bUCB Pharma S.A.,

Product Development, Braine l’Alleud B-1420, Belgium

* Correspondence: kalliopi.dodou@sunderland.ac.uk; Tel.: +44-0191-515-2503

Development of a small-scale spray-drying approach for amorphous

solid dispersions (ASDs) screening in early drug development

The present study details the development of a small-scale spray-drying approach

for the routine screening of amorphous solid dispersions (ASDs). This strategy

aims to overcome the limitations of standard screening methodologies like

solvent casting and quench cooling to predict drug-polymer miscibility of spray-

dried solid dispersions (SDSDs) and therefore to guarantee appropriate carrier

and drug-loading (DL) selection. A DoE approach was conducted to optimize

process conditions of ProCept 4M8-TriX spray-drying to maximize the yield

from a 100 mg batch of Itraconazole/HPMCAS-LF and Itraconazole/Soluplus

40:60 (w/w). Optimized process parameters include: inlet temperature, pump

speed, drying and atomizing airflows. Identified process conditions derived from

the DoE analysis were further i) tested with Itraconazole, Naproxen and seven

polymers, ii) adapted for small cyclone use, iii) downscaled to 20 mg batch

production. Drug-polymer miscibility was systematically characterized using

modulated differential scanning calorimetry (mDSC). Spray-drying was

identified as a well-suited screening approach: mean yield of 10.1 to 40.6% and

51.1 to 81.0% were obtained for 20 and 100 mg ASD productions, respectively.

Additionally, this work demonstrates the interest to move beyond conventional

screening approaches and integrate spray-drying during screening phases so that

a greater prediction accuracy in terms of SDSDs miscibility and performance can

be obtained.

Keywords: amorphous solid dispersions; spray-dryer; design of experiments;

screening; miscibility; solvent casting; quench cooling; polymers

Introduction

The increasing number of poorly water soluble compounds within drug pipelines poses

problems in the drug development strategy of orally administrated formulations

(Lipinski et al. 2012). The low solubility usually results in a limited dissolution rate and

reduces the oral bioavailability of the drug. In this regard, amorphous solid dispersions

(ASDs) can improve aqueous solubility and hence bioavailability of drugs (Vasconcelos

et al. 2007). This formulation strategy involves the dispersion of the amorphous drug

particle within a polymeric matrix, achieved by a melt or a solvent method (Teja et al.

2013). From an industrial perspective, spray-drying is a well-known process used to

convert solutions, suspensions and emulsions into powder at laboratory, pilot and

commercial scale (Paudel et al. 2013).

In the past decades, laboratory spray-dryers have been routinely used in the

pharmaceutical industry for the production of ASDs. Typically, this technique enables

the production of milligram to gram scale batches and thereby is particularly suitable to

support preclinical to early stage clinical activities (He and Ho 2015). In the current

study, the ProCept laboratory spray-dryer was used due to its capability to work with a

large range of feed solution volumes ranging from 0.5 mL to 24 liters/8 hours (ProCept

2014). Numerous studies have reported the optimization of spray-dried powder

properties with a predominant focus on yield and particle size distribution (Amaro et al.

2011; Schmid et al. 2011; Lebrun et al. 2012).

However, the ability of laboratory spray-drying to provide reliable productions

of solid dispersions during screening stage remains one major constraint (Ormes et al.

2013). It is well known that small batch size at the milligram scale results in a

significant yield reduction of the spray-dried material. Few studies have considered the

use of spray-drying for the production of solid dispersions at milligram scale (Chen et

al. 2014; Gu et al. 2015). Nevertheless, the majority of the formulations investigated

had generally limited drug-loading (DL) and were tested with a restricted number of

drug and excipients. To the best of our knowledge, little has been published on the

capability of spray-drying to be downscaled and adapted to the needs of screening

phases of solid dispersions in early drug development. The use of spray-drying during

small-scale ASDs screening remains limited in its current form and conventional

screening approaches, namely solvent casting and quench cooling are generally

preferred (Dai et al. 2008; Parikh et al. 2015).

In a recent study, the authors demonstrated that standard screening methods like

quench cooling and solvent casting performed at various evaporation rates cannot

guarantee appropriate carrier and DL selection due to their limited accuracy to predict

the phase behavior of spray-dried solid dispersions (SDSDs), consistently (Ousset et al.

2018). Despite conventional screening methodologies being the commonly used

approach in the pharmaceutical industry, the influence of the preparation method on the

properties and performance of ASDs generated during screening phase, and

consequently on the carrier selection, should not be neglected. Given the importance to

select appropriate polymer and DL in the early phase of drug development, there is an

interest to move beyond traditional screening approaches in order to better anticipate

SDSDs properties and performance. The novelty of this work consists of reconsidering

the use of spray-drying in a small-scale approach and improving the prediction accuracy

to determine the drug-polymer miscibility of SDSDs.

This work describes for the first time the development of a small-scale spray-

drying approach for the screening of binary ASDs at preclinical stage and aims to define

a generic method that can be used routinely in the pharmaceutical industry. In this

regard, a particular attention has been given to develop a method in line with the

screening requirements that: i) allows the testing of a large range of excipients and high

DL, ii) can be applied on a large range of active drugs including low glass transition

temperature (Tg) compounds, iii) produces solid dispersions at the milligram scale with

optimized yield, short development time and limited amount of raw material, iv)

respects the needs for subsequent analytical characterization of the screened

formulations, v) is representative and scalable with regard to formulation attributes and

characteristics as well as process parameters and conditions.

First, a design of experiments (DoE) approach was conducted to identify robust

processing conditions of drying airflow, inlet temperature, atomizing airflow and pump

speed in order to optimize the yield of 100 mg productions of Itraconazole ASDs. The

use of DoE is a well-established statistical method in compliance with quality by design

(QbD) principles, and encouraged by the authorities (Q8(R2)-ICH) to provide a better

understanding of the influence of formulation and process parameters on the quality

attributes of the final product (Lebrun et al. 2012; Kauppinen et al. 2017). Second, the

identified process conditions derived from the DoE approach were validated and tested

with more polymers (listed in Table 1), an additional drug (Naproxen) and different DL.

The process conditions were further adapted to the use of small cyclone, re-tested and

downscaled to assess the spray-drying capability. The phase behavior of solid

dispersions produced at small-scale ( ≤ 100 mg) was analyzed using modulated DSC

(mDSC) and X-ray powder diffraction (XRPD), and compared to samples prepared by

spray-drying at larger scale (2.5 g), and by standard screening methodologies, namely

solvent casting and quench cooling. Finally, the performance of screened SDSDs of

Itraconazole and Naproxen in terms of solubility enhancement and physical

stability was also investigated.

Materials and methods

Materials

Naproxen and Itraconazole were purchased from SRIS Pharmaceuticals (Hyderabad,

India). Hydroxypropylmethylcellulose phthalate (HPMCP HP50) and

hydroxypropylmethylcellulose acetate succinate fine grade (HPMCAS-LF) were

obtained from Shin-Etsu (Tokyo, Japan). Copolymer of N-vinyl-2-pyrrolidone and vinyl

acetate (PVPVA) was obtained from Ashland (Covington, KY, USA) and

polyvinylpyrrolidone (PVPK30) was purchased from VWR (Heverlee, Belgium).

Copolymer of polyvinyl caprolactam-polyvinyl acetate-polyethylene glycol graft co-

polymer (Soluplus) was donated by BASF (Ludwigshafen am Rhein, Germany).

Copolymer of methacrylic acid and methyl methacrylate 1:1 (Eudragit L100) and

copolymer of methacrylic acid and ethyl acrylate copolymer 1:1 (Eudragit L100-55)

were donated by Evonik (Essen, Germany). The solvents used were of analytical or

HPLC grade.

Methods

2.1 Generic method development

2.1.1 SDSD production

Spray-dried binary solid dispersions were prepared using the laboratory scale ProCept

4M8-TriX spray-dryer (Zelzate, Belgium). The feed solution was pumped to the nozzle

via a peristaltic pump Watson Marlow 530S (Falmouth, Cornwall, UK). The process

was operating with a length of two chambers and in the open loop configuration.

Cyclone was systematically used to separate solid particles from drying gas airflow via

centrifugal forces (Wang et al. 2006). Regarding this equipment, the supplier proposes

cyclone in four different sizes (small, medium, large, extra-large) for optimal product

recovery (ProCept 2014). Cyclone efficiency is directly impacted by the product

characteristics e.g. particle size and density (Miller and Gil 2012). In this study, medium

cyclone was initially used as it represents the best compromise approach to separate

particles with a wide range of size and operates with a wide range of processing

conditions. In this regard, medium cyclone is particularly adapted to the DoE approach

where variable process conditions are tested and therefore particles with different

properties are generated. Subsequently, identified process conditions derived from the

DoE approach were tested with small cyclone. Powder was collected into a 2 mL glass

vial for the 100 mg productions and into a standard aluminium pan (TA Instruments,

Leatherhead, UK) in the case of the 20 mg batch production. Powder collection was

done via a customized 3D printed funnel ensuring sealing of the system and reducing

material loss during powder handling. The collected powders were stored in a vacuum

oven for 48 hours. Output process parameters such as the chamber outlet temperature

and the pressure drop (∆P) over the cyclone were monitored via the acquisition system

interface.

2.1.2 DoE approach

Based on literature review and prior knowledge, a risk assessment was conducted

regarding yield impact and the following process parameters and material attributes

were identified: four process parameters (drying airflow, inlet temperature, atomizing

airflow and pump speed) and one material attribute (polymer type) (Patel BB et al.

2015; Singh and Van den Mooter 2016). As seen in Table 2, each process factor was

studied at three levels while the polymer type was included in the DoE as categorical

factor (two levels). Intrinsic properties of polymers such as Tg and viscosity were

covered via the polymer type factor. Additional key experimental factors such as nozzle

orifice diameter, cooling airflow, cyclone size, solid content, DL, model drug and

solvent were kept constant. The predefined values of each process and material input

are listed on Table 3.

In this DoE, the batch size and the DL were fixed at 100 mg and 40% (w/w),

respectively, which is assumed to be representative of the operating conditions during

screening at preclinical stage (Ayad et al. 2013; Lohani et al. 2014). Itraconazole, a

BCS class II compound known for its extremely low solubility in water (estimated at

approximately 1 ng/mL) was selected as model drug (Engers et al. 2010). Preliminary

tests (data not shown) prior to the DoE study were performed to evaluate the physico-

chemical properties of seven polymers (cellulose, PVP, acrylate and miscellaneous

polymers). This selection of seven polymers is a typical range of carriers tested

during screening phases. The use of enteric polymers during preclinical drug

development is of particular interest so that the limited hydration of such polymers

in acidic pH conditions is preventing the release of amorphous drug substance and

is limiting the risk of recrystallization into the stomach. The absorption can be

therefore maximized in intestinal fluids where polymer hydration occurs (Ueda et

al. 2014).

Analyses regarding Tg measurement and viscosity of polymer solution at various

concentrations were conducted due to their potential impact on the yield of spray-dried

material (Paudel et al. 2013). Based on this preliminary evaluation, Soluplus, a non-

ionic and amphiphilic polymer with relatively low Tg (approximately 70°C) (Djuris et

al. 2013) and HPMCAS-LF an enteric (pH 5.5) cellulosic derivative polymer known for

its relatively high Tg (approximately 122°C) and viscosifying properties at high

concentration (Ueda et al. 2014) were selected as they represented extreme properties

among excipients of the studied domain.

A response surface methodology (RSM) design including four continuous

factors (three levels) and one categorical factor (two levels) was developed using JMP

11 software (SAS, Cary, NC, USA). A total of 33 experiments including two repeated

experiments and three center points were performed (Appendix). The analysis of

variance (ANOVA) was computed to determine the statistical significance of each input

variable on the selected response (yield). The non-significant terms were removed from

the statistical model to optimize it. Then, the desirability of the response was evaluated

and factor values were identified at maximized yield desirability. Finally, two

dimensional contour plots were represented to visualize the influence of input variables

on the response (yield).

2.1.3 Yield calculation and specification

Yield was identified as the main quality attribute and was integrated as response in the

model for optimization purpose. Yield is calculated as the ratio between the amount of

particles collected and the total amount of solid dissolved in the feed solution. The yield

value is expressed in percentage (%).

Yield specifications were defined depending on the production size: more than

50% for 100 mg productions and more than 10% for 20 mg productions. The

specification limits would result in sufficient spray-dried material for subsequent

analytical characterization typically required during ASDs screening.

2.1.4 Model validation

In order to assess the robustness of the model, production of a 100 mg ASD batch of

Itraconazole/HPMCAS-LF and Itraconazole/Soluplus 40:60 (w/w) was performed in

triplicate using identified operating conditions and medium cyclone. The average yield

value was computed and compared to the 95% prediction interval provided by the

model.

2.1.5 Model verification and adaptation to the small cyclone use

Subsequently, the identified operating conditions were further: i) tested with the

production of solid dispersions of Itraconazole and Naproxen mixed with a set of seven

polymers at a fixed DL of 40% (w/w) (the listed parameters detailed in Table 3 were

kept constant), ii) adapted to the use of the small cyclone, iii) re-tested with the

productions of solid dispersions of Itraconazole and Naproxen at 40% and 20% (w/w)

DL using small cyclone. Naproxen was selected due to its challenging properties

regarding the manufacturing of low Tg compounds (approximately 6 °C). As

amorphous forms exist in a rubbery state above their Tg, the stickiness of

amorphous material to the wall of the column would reduce tremendously the

yield value of the spray-dried material (Paterson et al. 2005).

2.1.6 Downscaling

Optimized process conditions using small cyclone were further downscaled to ASD

production of 20 mg. Production of solid dispersions of Itraconazole and Naproxen

mixed with a set of seven polymers was performed. Each formulation was evaluated at a

DL of 20% and 40% (w/w), respectively.

2.2 Alternative ASD screening methodologies

2.2.1 Solvent casting

Stock solutions of drug and polymer with a solid content of 50 mg/mL were prepared

using a binary solvent mixture of DCM/EtOH 2:1 (v/v). The stock solutions were mixed

in the appropriate volumetric ratio and a final volume of 5 mL was spread on a teflon

plate (21 cm × 15 cm). A 15 cm diameter funnel (VWR, Heverlee, Belgium) was put on

top of the casting solution and the solvent evaporation was performed at room

temperature for 1 week. The film casted solid dispersions were then removed from the

teflon plate and were stored in a vacuum oven for 48 hours to remove residual solvent.

2.2.2 Quench cooling

Quench cooling screening was performed using a TA Instruments Q1000 DSC (TA

Instruments, Leatherhead, UK). Casted samples were heated at a temperature of up to

20 °C higher than the last thermal event e.g. melting of the drug or the Tg of polymer to

obtain a molten mixture. A fast cooling temperature of down to -50 °C was applied to

prevent the drug recrystallization from the molten state. Lastly, the quench cooled

samples were analyzed in mDSC.

2.3. Analytical methods

2.3.1 Modulated DSC

mDSC analyses were performed using TA Instruments Q1000 calorimeter (TA

Instruments, Leatherhead, UK). The chamber was purged with a 50 mL/min flow rate of

dry nitrogen. Indium was used for temperature and enthalpy calibration. The heat

capacity calibration was performed at 96.9 °C using sapphire disks. About 2 – 4 mg of

powder was analyzed in closed standard aluminium pans (TA Instruments, Leatherhead,

UK). Samples were heated from 0 °C to 210 °C at 2 °C/min combined with a

modulation of ± 1 °C and a period of 40 sec. The thermograms were collected and Tg

was evaluated in the reverse heat flow signal using Universal Analysis 2000 software

(TA Instruments, Leatherhead, UK).

2.3.2 X-ray powder diffraction

XRPD experiments were conducted on X Bruker AXS D8 Advance (Bruker, Karlsruhe,

Germany). A few milligrams of powder were dropped on the center of a silicium

monocrystal holder. Samples were analyzed over the range 4.5 - 30° at a scan speed of

2.5 sec/step and a step size of 0.02°. The diffractograms were collected and processed

using Eva DIFFRAC-SUITE software (Bruker, Karlsruhe, Germany). The integrated

patterns represented the intensity as a function of 2θ.

2.3.3 Scanning electron microscopy

Scanning electron microscopy (SEM) observations were performed using the

JEOL JSM-IT300 SEM (JEOL, Tokyo, Japan) to characterize the shape and

morphology of the spray-dried particles. Powder was attached to conductive

double-sided carbon adhesive tape mounted on an aluminium stud and coated with

gold. The SEM instrument was operated in high vacuum mode at an accelerating

voltage of 5 kV and at a working distance of 20.9 mm. The data treatment was

carried out using JEOL IT300 Operation software (JEOL, Tokyo, Japan).

2.3.4 Dissolution tests

Dissolution profiles of screened SDSDs of Itraconazole were obtained at a target

drug concentration of 1 mg/mL from a dissolution medium of 50 mM phosphate

buffer (pH 6.5) containing 2% Vitamin E TPGS as surfactant. Dissolutions tests of

screened SDSDs of Naproxen were performed in dissolution mediums consisting of

i) simulated gastric fluid without pepsin (pH 1.2) containing 0.5% SDS and 2%

HPMC at a drug concentration of 1 mg/mL, and of ii) 50 mM phosphate buffer

(pH 6.5) containing 0.5% SDS and 2% HPMC at 5mg/mL. Dissolution tests were

performed under non-sink conditions with respect to the crystalline drug. This

would allow discriminating screened SDSDs on their potential to generate and

maintain supersaturation (Sun et al. 2016). In this regard, target drug

concentration was selected greater than the solubility of crystalline drug in the

chosen dissolution mediums. Accurate weight of 2.5 mg and 12.5 mg of SDSD

(equivalent to 1 mg and 5 mg of drug, respectively) was distributed in 10 mL glass

tube (VWR, Heverlee, Belgium). 1 mL of cited above dissolution medium was

added to each tube. Temperature and magnetic stirring were maintained at 37°C

and 350 rpm, respectively, using Thermo Mixer C unit (Eppendorf, Hamburg,

Germany). During dissolution tests, samplings of 80 µL were withdrawn after 1, 5,

10, 15, 30, 60, 120 and 150 minutes. Samples were filtered on 0.45 µm ultrafree

centrifugal filter units (Merck Millipore, Burlington, MA, USA) and centrifugated

at 14000 rpm during 2 minutes. The filtrate was pipetted and then properly diluted

in H2O/ACN 1:1 (v/v). Itraconazole and Naproxen content was determined in

HPLC coupled by UV detection at 260 and 272 nm, respectively. Similar protocol

was applied to generate the dissolution profiles of crystalline Itraconazole and

Naproxen in the tested dissolution mediums. All experiments were performed in

triplicate.

2.3.5 Physical stability studies

Screened SDSDs of Itraconazole and Naproxen were stored at 40°C/75% RH and

25°C/60% RH for 5 weeks. Powder was analysed in XRPD at t0, t2weeks and t5weeks in

order to assess the potential of screened carriers to generate physically stable

SDSDs under both stress and ambient conditions.

Results and discussion

DoE approach

The results obtained for the DoE are detailed in Appendix. Yield values ranging from

19.2% to 70.7% were obtained for the 100 mg batch production of

Itraconazole/HPMCAS-LF and Itraconazole/Soluplus 40:60 (w/w). The model obtained

for the yield response was fitted using multiple linear regression. The statistical

significance of the factors was evaluated based on the confidence level of 95%. The

fitted model was adjusted by removing the non-significant factors while keeping the

main effects. The p-value of the ANOVA (< 0.05) and the adjusted R-Squared value of

0.76 show that the response is described significantly by the optimized model (Figure

1). Moreover, the model does not present lack of fit (p-value > 0.05), demonstrating its

suitability.

Based on DoE analysis and ANOVA results, seven factors including the main

effects and the interaction terms were identified to have a significant influence on the

response (p-values < 0.005). The main factors are ranked according to their impact on

yield: atomizing air > HPMCAS-LF formulations > inlet temperature > drying airflow >

interaction term between drying airflow and HPMCAS-LF formulations > pump speed

> interaction term between the drying airflow and the inlet temperature. Interestingly,

the type of polymer represented the second highest impacting effect on the yield. This

reveals that the influence of this material attribute is as important as process parameters

to understand yield variation of spray-dried material.

Figure 2 shows the conditions (factors values) identified for every polymer

formulation if the yield desirability is maximized. Mean yield values of 58.6% and

75.6% were predicted for solid dispersions of Itraconazole/HPMCAS-LF and

Itraconazole/Soluplus, respectively, at maximized desirability. The optimal values

identified for the drying (0.45 m3/min) and atomizing (1.5 bars) airflows as well as the

inlet temperature (60 °C) were the same for both HPMCAS-LF and Soluplus ASDs.

The optimized value of the inlet temperature defined by the model was found at

the lowest level of 60 °C. Indeed, the yield was negatively impacted by an increased

inlet temperature. Low inlet temperature reduces the outlet temperature at the bottom of

the chamber and at the entry of the cyclone (Cal and Sollohub 2010). Under the

optimized process parameter conditions, the outlet temperature was measured at around

40 °C; the outlet temperature is a good indicator of the product temperature to which the

dried droplets are exposed (Maas et al. 2011; Paudel et al. 2013). High outlet

temperature close or above the Tg of amorphous products favors material sticking on the

glass walls and hence reduces significantly the yield (Paterson et al. 2005). Therefore, a

low outlet temperature is particularly suitable for the production of ASDs with low Tg

drug and/or polymer.

According to the model, drying and atomizing airflows were positively

correlated to the yield (Figure 2). Interestingly, atomization airflow was found as the

main effect impacting the response and was positively correlated to the yield variation.

In addition, a value of 0.45 m3/min of drying airflow was found to maximize the yield

for both formulations. This observation is consistent with the theory of spray-drying:

high drying airflow is reducing the residual moisture of the final product and improves

the particle separation in the cyclone (Maury et al. 2005). The value of ∆P over the

cyclone represents the cyclone efficiency to separate particles from the drying gas

(Elsayed and Lacor 2011). The ∆P value is mostly influenced by the drying and the

cooling airflows. A ∆P value of 51 mbar was measured during the tests with the

optimized method. This value was in agreement with the supplier specification to ensure

an optimal particle separation (ProCept 2014).

Yet, pump speed was found to have a limited influence on the yield; DoE

analysis revealed that the pump speed was identified as the least significant factor

among the main effect terms. Moreover, the pump speed was found to influence the

yield in an opposite way depending on the used polymer. The pump speed has a slightly

negative effect on the yield of HPMCAS-LF ASDs but a slightly positive effect on the

yield of Soluplus ASDs. This difference can be attributed to feed solution properties

such as viscosity which is inherently impacted by the polymer type.

Yield distribution map was constructed as a function of pump speed with each

other variable process parameter. This aims to determine the pump speed value that

enables the production of ASDs with various polymers in the context of generic method

development. Figure 3 displays the two dimensional contour plot results that represent

yield map for productions of 100 mg ASD of Itraconazole/HPMCAS-LF and

Itraconazole/Soluplus 40:60 (w/w). The contour plots showed that a low pump speed in

combination with optimal process parameter for inlet temperature, drying and atomizing

airflows (Table 4 for further details) led to a maximum yield. Consequently, the pump

speed value was set at 1 g/min.

Furthermore, the fitted model was used to simulate the average yield of 5000

productions using optimized process parameters as described in Table 4. The

simulations were computed and the results are reported in Figure 4. Based on these

simulations, the model predicted a mean yield above the specification limits being in

line with minimum quantities needed for further characterization. Regarding HPMCAS-

LF formulations, the simulations predict a risk of 5.3% to obtain a yield lower than 50%

which can be considered as an acceptable risk in the scope of the development of a

generic method for preclinical screening of ASDs.

Model validation

Production of 100 mg ASD batches of Itraconazole/HPMCAS-LF and

Itraconazole/Soluplus 40:60 (w/w) was performed in triplicate under optimized process

conditions using medium cyclone. Mean yield values of 54.0% ± 8.4 and 56.6% ± 3.6

were experimentally obtained for solid dispersions of Itraconazole/HPMCAS-LF and

Itraconazole/Soluplus, respectively, while the prediction interval at 95 % provided by

the model predicted a yield of 47.9 to 69.1% and of 61.9 to 83.7%, respectively. The

prediction interval provided by the model overestimates the experimental results

generated for the production of Soluplus ASDs. However, experimental yield results

were found above the specification limits for both mixtures. The last observation

confirms the selection of identified process conditions in line with the scope of the

study.

One explanation to the differences obtained between predicted and experimental

yield results is the lack of understanding of response by the fitted model. The adjusted

R-Squared value of the model (0.76) attests that other parameters influencing the yield

need to be taken into consideration to better explain the observed variability (Figure 1).

Concentration effects and solvent choice have not been investigated in this study and

would probably impact the response too (Paudel et al. 2013).

Model and/or process parameter verification

The suitability of the optimized manufacturing method was then evaluated by producing

a 100 mg batch of Itraconazole and Naproxen ASDs using medium cyclone. Samples

were prepared in triplicate at 40% (w/w) DL with a set of seven polymers listed in

Table 1. Table 5 reports the average yield and standard deviation values experimentally

obtained. Mean yield values ranging from 44.4 to 56.3% and 49.0 to 59.6% were

obtained for Naproxen and Itraconazole ASDs, respectively.

Nevertheless, four Naproxen/polymer and one Itraconazole/polymer

combinations did not reach the specification value of 50% as shown in Table 5. The

latter observation suggests that the initial choice to select HPMCAS-LF and Soluplus

(mainly based on their thermal and viscosity properties) as model carriers in the DoE

design may have been too restricted and other material attributes may have been omitted

(density, surface tension). The identified process conditions in their current form are not

fitting to these new drug/polymer combinations and to a large extent cannot be applied

as screening procedure as the yield specification has not been reached consistently.

Adaptation to the small cyclone use and model and/or process parameter

verification

The identified operating conditions were further adapted to the use of the small cyclone:

the drying airflow was decreased to 0.3 m3/min to ensure a similar outlet temperature

and a similar pressure drop over the cyclone compared to the previous operating

conditions with the medium cyclone. As the small cyclone can only operate in a

restricted domain, lower drying airflow was applied to avoid pressure overload in the

system. All other parameters listed in Table 3 were kept constant. Table 4 summarizes

the operating conditions applied using small cyclone.

Productions of 100 mg batches of Itraconazole and Naproxen ASDs were

repeated under these novel operating conditions. SEM microphotographs of 40:60

(w/w) Itraconazole SDSDs collected from the small cyclone are displayed in Figure

5. Particles smaller than 10 µm were obtained and collected after processing. The

spray-dried particles showed spherical shape with both smooth and shrinked

surface (Figure 5 a-b) and shriveled surface (Figure 5 c-f). This morphological

shape is typical for spray-dried material manufactured at low inlet temperature

(Tonon et al. 2008). Moreover, the structural morphological differences between

powder made with different carriers arise from the physical characteristics of the

solid crust formed during solvent evaporation which is largely influenced by

polymer type (Vicente et al. 2013).

Yield values obtained for the productions of ASD 40:60 (w/w) are detailed in

Figure 6. Under these conditions, all screened drug-polymer combinations resulted in a

yield of above 50%. The results obtained are mainly explained by the improved

performance of the small cyclone to collect powder from drying air. In the present

study, the highest selectivity of small cyclone is particularly adapted in the case of the

identified process conditions derived from the DoE approach: where the combination of

a high atomizing gas flow (1.5 bars) and a low pump speed (1 g/min) should

theoretically favor the formation of small particle sizes. Application of high spray gas

flow will produce smaller droplets and hence smaller particle. At a constant atomizing

gas flow, a lower pump speed will decrease the droplet size and hence the particle size

(Patel BB et al. 2015). In this regard and based on the supplier documentation, the small

cyclone offers the most selective particle separation of up to 1.5 µm (ProCept 2014).

This finding contrasts with previous studies where a large particle size correlates with a

high yield (He and Ho 2015).

Figure 7 represents the yield improvement of ASDs produced using small

cyclone compared to initial medium cyclone. The use of the small cyclone improved the

yield of the 100 mg ASD batches for both drugs, significantly. The overall yield for

ASDs of Naproxen was 51.0% with the medium cyclone and 65.9% with the small

cyclone. Similarly, the overall yield of solid dispersions of Itraconazole was improved

from 54.4% to 68.0% by reducing cyclone size. Despite a significant increase in yield

compared to the medium cyclone, the yield improvement was found to depend on the

nature of the dug - polymer combination (Figure 7). This finding confirms earlier

observations and highlights the importance of material attributes and more specifically

feed solution properties on the response. Yield improvement was lower for solid

dispersions made with HPMCP HP50, HPMCAS-LF, Eudragit L100 and Eudragit

L100-55. These polymers were found to have the most important viscosifying

properties among the set of polymers investigated (data not shown). Feed solutions with

high viscosity will engender bigger particles (Davis et al. 2017), as a result the relative

selectivity of the small cyclone compared to the medium size is probably be the least

significant.

Figure 6 represents the results obtained for additional productions of

Itraconazole and Naproxen ASD at 20% (w/w) under optimized process conditions

using the small cyclone. Yield results obtained at 20% DL (w/w) were found above the

specification limits, consistently. Comparable yields were obtained for most of the drug-

polymer combinations with a DL of 20% and 40% (w/w) except for

Itraconazole/PVPVA and Naproxen/Soluplus. In these particular cases, the yield was

decreased from 40% to 20% (w/w) DL. The last observation contradicts the common

assumption that reducing DL is expected to increase the yield of spray-dried material,

theoretically. This is because the Tg value of ideal glass solution is function of the Tg of

the pure components in the blend and of the system composition: a reduction in DL is

increasing the mixing Tg due to higher polymer content (Teja et al. 2013).This

assumption considers the fact that generally the Tg of the polymer is higher than the Tg

of the drug (Table 1). This argument takes into account ideal glass solutions, only. In

this regard, amorphous forms with high Tg are usually considered as suitable for spray-

drying production due to the limited risk of stickiness on the glass wall (Paterson et al.

2005). Nevertheless, results obtained in the present study highlighted that the nature of

the drug and its inherent properties such as Tg did not influence the yield. Mean yield of

20% and 40% (w/w) solid dispersions of Naproxen and Itraconazole was 64.1 ± 4.7%

and 66.8 ± 5.6%, respectively. Consequently, the proposed method can be considered as

a well-suited approach for the ASDs screening of low Tg drug such as Naproxen.

Downscaling

Downscaling trials were performed to assess the capability of spray-drying to operate at

minimum batch size while supporting the requirements for screening and analytical

characterization. At this stage, a yield superior to 10% corresponding to 2 mg of ASD

collected is sufficient to investigate the phase behavior and the solid state of screened

formulations. As shown in Figure 8, optimized process conditions using the small

cyclone were tested for 20 mg batch productions of Itraconazole and Naproxen ASD at

20 and 40% DL (w/w), respectively. Mean yield values above the specification limits

were obtained for all screened drug-polymer mixtures under these operating conditions.

The lowest yield, 18.2 ± 8.1% corresponding to an average mass of 3.6 ± 1.6 mg of

ASD was obtained for Naproxen/PVPK30 40:60 (w/w), whereas, the highest yield, 30.1

± 10.5%, corresponding to a mass of 6.0 ± 2.1 mg of ASD was found for

Itraconazole/PVPK30 40:60 (w/w).

In general, the results obtained at 40% (w/w) DL for Itraconazole samples were

slightly above the results found for Naproxen samples except for HPMCP HP50. Data

obtained at a batch size of 20mg suggest, that the yield seems to depend on the nature of

the drug and the involved polymer; for instance, the yield was positively impacted by a

lower DL for Naproxen samples mixed with HPMCAS-LF, PVPVA and Soluplus.

However, similar yields were obtained for other Naproxen ASDs at both DL. Likewise

for ASDs containing Itraconazole, the DL did not impact the yield and no correlation

between the yield and the DL was found at a 20 mg scale. As the spray-drying is

running at lowest possible operating conditions, the differences that are expected are

probably hindered by these extreme operating conditions. Additional downscaling trials

with a batch size of 10 mg were performed (data not shown). However, the minimal

yield of 2 mg was not achieved for all drug-polymer systems, consistently.

Drug-polymer miscibility characterization

Along with the development of a generic method adapted for the screening of small size

batches of ASD, the drug-polymer miscibility and the solid state of each SDSD

produced at small and larger scale (20, 100 mg and 2.5 g) was characterized using

mDSC and XRPD. Process and formulation parameters fixed during this DoE approach

and listed in Table 3, were kept constant for the larger scale productions. Moreover, the

following conditions of pump speed (6 g/min), inlet temperature (65 °C), drying airflow

(0.35 m3/min) and atomization airflow (1 bar) were used.

Figure 9 displays the mDSC thermograms of Itraconazole/PVPVA 40:60 (w/w)

produced by spray-drying at various scales and prepared by standard screening

methodologies, namely solvent casting and quench cooling. As seen in Figure 9, a glass

solution, depicted by the presence of a single Tg in the reverse heat flow signal was

obtained for the quench cooled sample. Residual crystallinity was detected in the film

casted sample by the presence of the drug melting endotherm in both reverse and total

heat flows. Additionally, the Tg of pure glassy Itraconazole with its inherent mesophase

endotherms (Six et al. 2001) was found in the thermogram of film casted ASD and

suggests the presence of a phase separated system (Six et al. 2002). Samples produced

by spray-drying were identified as solid glass suspension. These samples are

characterized by the presence of the Tg of the pure amorphous drug, the inherent

mesophases of glassy Itraconazole and the Tg of the polymer. The absence of residual

crystallinity was confirmed in the total heat flow signal of mDSC and by XRPD (data

not shown). These results confirm that the above commonly used screening methods

cannot predict the phase behavior of SDSDs.

The prediction accuracy of our small-scale spray-drying approach and

conventional screening methodologies to determine the drug-polymer miscibility of

SDSDs is compared Table 6, through the entire set of ASD productions giving a total of

28 samples. As seen in Table 6, the highest accuracy was obtained in the case of small-

scale spray-drying screening of 20 and 100 mg: these approaches allow for the drug-

polymer miscibility prediction of 27/28 and 28/28 of SDSDs tested, respectively. One

difference was obtained for Itraconazole/Soluplus 20:80 (w/w) produced at 20 mg and

identified as a glass suspension while drug recrystallization occurred during analysis at

100 mg and 2.5 g. On the contrary, solvent casting and quench cooling did not provide a

reliable insight into the prediction of the phase behavior of SDSDs over the set of tested

drug-polymer combinations. Lower accuracy to predict the phase behavior of SDSDs

than small-scale spray-drying approach was found for solvent casting and quench

cooling with predictive ratio of 16/28 and 18/28 of SDSDs tested, respectively. In

addition, the evaluation of thermograms (data not provided) showed that similar Tg

value was obtained for samples produced by spray-drying at various scales. As Tg is

known to provide insight into the stability and homogeneity of amorphous material, this

demonstrated the superior ability of spray-drying screening compared to standard

methodologies to anticipate the final properties and performance of SDSDs (Engers et

al. 2010). This confirms the strategy developed in the current study in order to

overcome the limitations of standard screening tests to predict SDSDs properties and

performance in early phase of drug development.

Assessment of screened SDSDs potential: insight into physical stability and

dissolution performance

Carrier performance with regard to the manufacture of glass solutions that

improve drug solubility and that remain physically stable upon storage needs to be

assessed during the screening phase (Chiang et al. 2012). In this regard, the

performance of 100 mg batches of Itraconazole and Naproxen 40:60 (w/w) SDSDs

have been investigated in terms of physical stability and dissolution performance.

The selected drug-polymer ratio is appropriate to reach the high doses to be tested

during preclinical stage of drug development in the pharmaceutical industry

(Lohani et al. 2014).

The XRPD patterns of 40:60 (w/w) SDSDs of Itraconazole are depicted in

Figure 10. All screened drug-polymer systems were characterized by the presence

of amorphous halo. The absence of peaks relative to crystalline drug in XRPD

pattern confirms that all screened samples are manufactured in amorphous state

after processing. However, as aforementioned in previous section, the evaluation of

drug-polymer miscibility using mDSC confirms the presence of glass solution

system with HPMCP HP50, HPMCAS-LF, Eudragit L100 and Eudragit L100-55

and the presence of phase separation with PVPVA and PVPK30, as seen in Table

6. Furthermore, glass solution followed by drug recrystallization during

measurement was obtained in the case of Itraconazole/Soluplus system. Therefore,

mDSC was proven to be a more sensitive technique to differentiate stable glass

solutions from those which are prone to partial phase separation and drug

recrystallization during the heating process (Baird and Taylor 2012).

The physical stability of Itraconazole-polymer combinations was

investigated under stress (40°C/75% RH) and ambient (25°C/60% RH) storage

conditions. Results are summarized in Table 7. Under standard storage conditions,

all Itraconazole SDSDs maintained their inherent amorphous state during 5 weeks

at 25°C/60% RH. XRPD pattern of Itraconazole SDSDs stored at 40°C/75% RH

for 5 weeks are shown in Figure 10. Under stress conditions, Itraconazole systems

made with HPMCP HP50, HPMCAS-LF, Eudragit L100 and Eudragit L100-55

were found to remain amorphous. However, drug recrystallization was confirmed

for Itraconazole/Soluplus system after 2 weeks at 40°C/75% RH. This finding

corroborates the thermal signature of Itraconazole/Soluplus obtained in mDSC

where drug recrystallization was recorded during analysis. Indeed, drug

recrystallization during the heating procedure of mDSC is a sign of limited

capacity of polymer to stabilize the amorphous drug and therefore can provide

valuable insight into the physical stability of amorphous system (Duarte et al.

2015). As seen in Figure 10, the onset of drug recrystallization detected for solid

dispersions of Itraconazole with PVPVA and PVPK30 stored 5 weeks under stress

conditions confirms the fact that phase separated systems are more prone to

recrystallization than glass solutions.

Figure 10 displays the XRPD pattern of 40:60 (w/w) SDSDs of Naproxen

after manufacturing. Examination of XRPD patterns confirms that only ASDs

made with PVPVA and PVPK30 maintained a complete amorphous state after

processing. Evidence of drug residual crystallinity was observed for remaining

SDSDs made with HPMC HP50, HPMCAS-LF, Soluplus, Eudragit L100 and

Eudragit L100-55. This finding confirms drug-polymer miscibility as summarized

in Table 6. In this regard, only Naproxen-polymer systems characterized in

amorphous state i.e. PVPVA and PVPK30 ASDs were subjected to stability

studies. Results from stability study are summarized in Table 7 and XRPD pattern

of Naproxen ASDs stored at 40°C/75% RH during 5 weeks is depicted in Figure

10. Under both ambient and stress conditions, Naproxen/PVPVA and

Naproxen/PVPK30 were found to maintain complete drug amorphous state up to 5

weeks, which is a good indicator of polymer potential to stabilize amorphous drug

upon storage. This argument strengthens the selection of such carriers for

Naproxen SDSD manufacturing. Storage of ASDs systems containing hygroscopic

carriers such as PVPVA and PVPK30 should be ideally done under dried

conditions to reduce drug mobility and hence prevent drug recrystallization

(Rumondor et al. 2009).

In parallel to stability studies, the dissolution performance of screened

Itraconazole SDSDs was assessed at 37°C in dissolution medium at pH 6.5

containing surfactants. Dissolution profile of screened Itraconazole SDSDs and

solubility improvement percentage compared to crystalline drug after 60 and 150

minutes are depicted in Figure 11. After 60 minutes, all screened ASDs allowed to

generate supersaturation and improve drug solubility, significantly. Solubility

improvement in the range of 5679-6931% was obtained and thus confirms the

greater solubility of amorphous Itraconazole form compared to crystalline

counterpart. After 150 minutes, only four ASDs made with HPMCAS-LF,

Soluplus, Eudragit L100 and Eudragit L100-55 allowed to maintain

supersaturation during dissolution tests. These dispersions systems display similar

solubility improvement percentages between 60 and 150 minutes. However,

recrystallization process during dissolution tests characterized by sudden drop in

solubility value was recorded for ASDs containing HPMCP HP50, PVPVA and

PVPK30. Indeed, such ASDs lost their potential in enhancing drug solubility. As

an example, solubility improvement of Itraconazole/PVPK30 fell back to 263%

after 150 minutes.

The obtained dissolution profile of screened Itraconazole SDSDs correlates

well with their inherent solid state and drug-polymer miscibility characterization

discussed in the previous section. Indeed, ASDs containing HPMCAS-LF, Eudragit

L100 and Eudragit L100-55, previously identified as glass solutions, were able to

generate and maintain supersaturation during the entire dissolution test.

Furthermore, Itraconazole/Soluplus was also found to provide improved drug

solubility up to 150 minutes during the dissolution tests. However, as seen in Table

7, this system tends to recrystallize upon storage at 40°C/75% RH. In addition,

dissolution tests (data not shown) performed in the case of Itraconazole/Soluplus

stored during 2 weeks at 40°C/75% RH reveal that this ASD lost its potential in

increasing drug solubility contrary to other glass solutions: solubility improvement

of 5679/5347% and 1097/825% at 60/150 minutes were obtained for

Itraconazole/Soluplus after manufacturing and after 2 weeks under stress

conditions, respectively. As aforementioned, phase separation was observed during

solid state characterization of SDSDs containing PVPVA and PVPK30. This

translated well with their inherent dissolution profile where recrystallization

process was observed during dissolution tests. This demonstrates the higher

tendency of phase separated system to revert back to their crystalline form during

both dissolution tests and physical stability (Huang and Dai 2014). Interestingly,

Itraconazole/HPMCP HP50 was found to not maintain supersaturation during the

length of dissolution test, despite this system was identified as physically stable

amorphous glass solution under stress and ambient conditions. This finding

suggests that the mechanism of polymer stabilization that occurs at solid state and

during dissolution is different (Brouwers et al. 2009; Chauhan et al. 2013) which

justified that both dissolution and physical stability performance of ASD should be

addressed during screening phases. The selection of HPMCP HP50 as potential

carrier for Itraconazole SDSD development would not be considered as it does not

meet the key expectations of solid dispersion performance during screening

phases.

Similarly, the potential of screened SDSDs of Naproxen to improve drug

solubility was investigated. Because Naproxen is an acidic compound of which

solubility varies greatly depending on pH value, dissolutions tests have been

performed in dissolution mediums representative to pH of both gastric and

intestinal fluids (Chowhan 1978). First, solubility improvement results of screened

SDSDs of Naproxen recorded after 60 and 150 minutes in dissolution medium at

pH 6.5 containing surfactants, are depicted in Figure 12. Only three screened

systems containing PVPVA, PVPK30 and Soluplus were found to generate and

maintain supersaturation during dissolution test. On the contrary, SDSDs made

with HPMCP HP50, HPMCAS-LF, Eudragit L100 and Eudragit L100-55

recrystallized in the fists seconds of dissolution tests (data not shown) which

explains the fact that solubility was not improved even after 60 minutes. As seen in

Table 6, the fact that residual crystallinity was observed for these specific

Naproxen-polymer systems after processing translates well with their limited

dissolution potential. Likewise, Naproxen/PVPVA and Naproxen/PVPK30,

previously identified as glass solution were found to display the best dissolution

performance with solubility improvement in the range of 50% after 150 minutes.

Interestingly, Naproxen/Soluplus exhibited similar potential that ASDs containing

PVPVA and PVPK30 in terms of solubility enhancement, despite the fact that

complete amorphization was not achieved after processing as seen in Figure 10. In

this context, Thakral et al. found that Camptothecin/Soluplus solid dispersion led

to a 75-fold increase of drug solubility while incomplete amorphization was

reported after processing (Thakral et al. 2012). Potential explanations may arise

from the amphiphilic structure of Soluplus which has been demonstrated to

solubilize drug and retard crystallization mechanisms through micellar formation

(Shamma and Basha 2013; Patnaik 2016). Herein, Naproxen-Soluplus affinity

would probably help in the stabilization of supersaturated solution generated by

dissolved amorphous drug fraction and prevent additional recrystallization

process during dissolution test.

The potential of PVPVA, PVPK30 and Soluplus to improve Naproxen

solubility was confirmed in gastric dissolution medium at pH 1.2 containing

surfactants. Figure 13 displays the dissolution profile and solubility improvement

percentage compared to crystalline drug after 60 and 150 minutes of such

Naproxen SDSDs. All drug-polymer combinations were found to generate and

maintain supersaturation during dissolution tests. Comparable solubility

improvement percentages ranging from 203-257% were obtained among tested

carriers after 60 minutes. Soluplus based solid dispersion was found to provide

similar dissolution performance than identified glass solutions. Additional tests

including biorelevant conditions and simulation of intestinal absorption would be

needed to discriminate among screened systems (Linn et al. 2012; Puppolo et al.

2017). Despite this, PVPK30 and PVPVA were found to offer the best guarantee to

achieve the manufacturing of physically stable amorphous system of Naproxen

upon storage and providing solubility enhancement.

The results obtained in the present study prove that spray-drying can be adapted

during the screening phases of ASDs and replaces existing screening methods.

Productions of 100 mg of ASD under optimized operating conditions would provide

enough material to ensure conventional preclinical screening activities (solid state

characterization, dissolution tests, physical stability assessment). In parallel,

downscaled trials allowed to assess the capability of spray-drying to run at lowest

possible operating conditions while supporting the needs for screening and analytical

characterization. In this regard, identified processing conditions allow the production of

solid dispersion of 20 mg batch size that can be used as first step in carrier selection

with regards to solid state characterization of screened samples (mDSC, XRPD,

polarized light microscopy…).

Otherwise, the proposed screening methodology focused on preclinical and

small-scale manufacturing activities with a DL investigated up to 40% (w/w).

Nevertheless, high DLs are commonly preferred in the final drug formulation regarding

late clinical studies (Demuth et al. 2015; Patel S et al. 2017). In this regard, the

proposed method would need to be further tested with higher DL to confirm that the

amount of material collected is sufficient to cover the needs during screening phase.

Additionally, a particular focus was placed on the ability of the proposed small-scale

approach to predict the phase behavior of SDSDs which was found to provide valuable

information regarding the physical stability and dissolution performance of screened

SDSDs. Nevertheless, particle size distribution has not been particularly investigated in

this study: it is assumed that particle size of powder generated at small-scale would not

be representative for material produced by larger scale equipment where the nozzle

geometry would generate larger particle size (Cal and Sollohub 2010).

In this study, DoE approach has been chosen to optimize process conditions of

spray-dried material. As mentioned in the previous section, this approach has limitations

such as the lack of understanding of response by the fitted model. However, when used

in conjunction with mechanistic modelling can lead to a better understanding of the

optimization process (Van Daele et al. 2017; Van Bockstal et al. 2018).

Conclusions

This study investigated the development of a generic screening method for ASD based

on a small-scale spray-drying approach that respects the requirements of screening

activities in the pharmaceutical industry. Optimized process conditions derived from

DoE approach allowed the consistent production of ASD batches of 100 and 20 mg and

would ensured the production of sufficient material to support analytical

characterization during screening phases. The proposed method was found particularly

suitable for the testing of a large number of carriers and drugs, including low Tg

compounds, and up to 40% (w/w) DL. In practical terms, this approach respects the

scope of screening methodologies in particular the limited drug supply in early drug

development: 4 to 8 mg and 20 to 40 mg of drug were needed per drug-polymer

combination for the production of batch size of 20 mg and 100 mg, respectively.

Although solvent casting and quench cooling approaches are known to operate in

automated way and limit loss of product during screening phases, our proposed method

constitutes a reliable alternative as it achieved the best trade-off regarding drug

consumption, work flexibility and short processing time.

The main benefit of the implementation of spray-drying during screening phases

is the improved prediction accuracy in terms of drug-polymer miscibility and

performance of SDSDs: miscibility prediction accuracy of 27/28 and 28/28 of SDSDs

tested was found for 20 and 100 mg SDSDs screening, respectively while lower

accuracy was obtained for solvent casting (16/28) and quench cooling (18/28). In that

respect, spray-drying screening was found to be representative and scalable with regards

to process parameters as well as formulation attributes. This novel approach would

allow to erase errors in polymer and DL selection during screening phases, ease the

transfer from screening phase to laboratory batch production, and shorten delivery

deadlines in the pharmaceutical industry. The outcome of the work demonstrates the

interest to move beyond standard screening approaches and to consider the use of spray-

drying in early phase of drug development so that a better prediction of SDSDs

properties and performance can be achieved. Indeed, this approach demonstrated

that screened SDSDs of Itraconazole and Naproxen could be accurately

discriminated in terms of physical stability and dissolution properties.

Acknowledgements, the authors would like to thank the PhD grant from the Product

Development department, the technical support and assistance from the laboratory service and

the solid state characterization group of UCB Pharma.

Declaration of interest, the authors report no conflict of interest.

References

Amaro MI, Tajber L, Corrigan OI, Healy AM. 2011. Optimisation of spray drying

process conditions for sugar nanoporous microparticles (NPMPs) intended for

inhalation. Int J Pharm. 421(1):99-109.

Ayad MH, Bonnet B, Quinton J, Leigh M, Poli SM. 2013. Amorphous solid dispersion

successfully improved oral exposure of ADX71943 in support of toxicology

studies. Drug Dev Ind Pharm. 39(9):1300-1305.

Baird JA, Taylor LS. 2012. Evaluation of amorphous solid dispersion properties using

thermal analysis techniques. Adv Drug Deliv Rev. 64(5):396-421.

Brouwers J, Brewster ME, Augustijns P. 2009. Supersaturating drug delivery systems:

the answer to solubility-limited oral bioavailability? J Pharm Sci. 98(8):2549-

2572.

Cal K, Sollohub K. 2010. Spray Drying Technique. I: Hardware and Process

Parameters. J Pharm Sci. 99(2):575-586.

Chauhan H, Hui-Gu C, Atef E. 2013. Correlating the behavior of polymers in solution

as precipitation inhibitor to its amorphous stabilization ability in solid

dispersions. J Pharm Sci. 102(6):1924-1935.

Chen XQ, Stefanski K, Shen H, Huang C, Caporuscio C, Yang W, Lam P, Su C,

Gudmundsson O, Hageman M. 2014. Oral delivery of highly lipophilic poorly

water-soluble drugs: spray-dried dispersions to improve oral absorption and

enable high-dose toxicology studies of a P2Y1 antagonist. J Pharm Sci.

103(12):3924-3931.

Chiang PC, Ran Y, Chou KJ, Cui Y, Sambrone A, Chan C, Hart R. 2012. Evaluation of

drug load and polymer by using a 96-well plate vacuum dry system for

amorphous solid dispersion drug delivery. AAPS PharmSciTech. 13(2):713-722.

Chowhan ZT. 1978. pH-Solubility Profiles of Organic Carboxylic Acids and Their

Salts. J Pharm Sci. 67(9):1257-1260.

Dai WG, Pollock-Dove C, Dong LC, Li S. 2008. Advanced screening assays to rapidly

identify solubility-enhancing formulations: high-throughput, miniaturization and

automation. Adv Drug Deliv Rev. 60(6):657-672.

Davis MT, Potter CB, Mohammadpour M, Albadarin AB, Walker GM. 2017. Design of

spray dried ternary solid dispersions comprising itraconazole, soluplus and

HPMCP: Effect of constituent compositions. Int J Pharm. 519(1-2):365-372.

Demuth B, Nagy ZK, Balogh A, Vigh T, Marosi G, Verreck G, Van Assche I, Brewster

ME. 2015. Downstream processing of polymer-based amorphous solid

dispersions to generate tablet formulations. Int J Pharm. 486(1-2):268-286.

Djuris J, Nikolakakis I, Ibric S, Djuric Z, Kachrimanis K. 2013. Preparation of

carbamazepine-Soluplus solid dispersions by hot-melt extrusion, and prediction

of drug-polymer miscibility by thermodynamic model fitting. Eur J Pharm

Biopharm. 84(1):228-237.

Duarte I, Santos JL, Pinto JF, Temtem M. 2015. Screening methodologies for the

development of spray-dried amorphous solid dispersions. Pharm Res. 32(1):222-

237.

Elsayed K, Lacor C. 2011. The effect of cyclone inlet dimensions on the flow pattern

and performance. Appl Math Model. 35(4):1952-1968.

Engers D, Teng J, Jimenez-Novoa J, Gent P, Hossack S, Campbell C, Thomson J,

Ivanisevic I, Templeton A, Byrn S et al. 2010. A solid-state approach to enable

early development compounds: selection and animal bioavailability studies of an

itraconazole amorphous solid dispersion. J Pharm Sci. 99(9):3901-3922.

Gu B, Linehan B, Tseng YC. 2015. Optimization of the Buchi B-90 spray drying

process using central composite design for preparation of solid dispersions. Int J

Pharm. 491(1-2):208-217.

He Y, Ho C. 2015. Amorphous Solid Dispersions: Utilization and Challenges in Drug

Discovery and Development. J Pharm Sci. 104(10):3237-3258.

Huang Y, Dai WG. 2014. Fundamental aspects of solid dispersion technology for

poorly soluble drugs. Acta Pharm Sin B. 4(1):18-25.

Kauppinen A, Broekhuis J, Grasmeijer N, Tonnis W, Ketolainen J, Frijlink HW,

Hinrichs WLJ. 2017. Efficient production of solid dispersions by spray drying

solutions of high solid content using a 3-fluid nozzle. Eur J Pharm Biopharm.

123:50-58.

Lebrun P, Krier F, Mantanus J, Grohganz H, Yang M, Rozet E, Boulanger B, Evrard B,

Rantanen J, Hubert P. 2012. Design space approach in the optimization of the

spray-drying process. Eur J Pharm Biopharm. 80(1):226-234.

Linn M, Collnot EM, Djuric D, Hempel K, Fabian E, Kolter K, Lehr CM. 2012.

Soluplus(R) as an effective absorption enhancer of poorly soluble drugs in vitro

and in vivo. Eur J Pharm Sci. 45(3):336-343.

Lipinski CA, Lombardo F, Dominy BW, Feeney PJ. 2012. Experimental and

computational approaches to estimate solubility and permeability in drug

discovery and development settings. Adv Drug Deliv Rev. 64:4-17.

Lohani S, Cooper H, Jin X, Nissley BP, Manser K, Rakes LH, Cummings JJ, Fauty SE,

Bak A. 2014. Physicochemical properties, form, and formulation selection

strategy for a biopharmaceutical classification system class II preclinical drug

candidate. J Pharm Sci. 103(10):3007-3021.

Maas SG, Schaldach G, Littringer EM, Mescher A, Griesser UJ, Braun DE, Walzel PE,

Urbanetz NA. 2011. The impact of spray drying outlet temperature on the

particle morphology of mannitol. Powder Technol. 213(1-3):27-35.

Maury M, Murphy K, Kumar S, Shi L, Lee G. 2005. Effects of process variables on the

powder yield of spray-dried trehalose on a laboratory spray-dryer. Eur J Pharm

Biopharm. 59(3):565-573.

Miller DA, Gil M. 2012. Chapter Spray-Drying Technology, Spray-Drying Technology.

In: Williams III RO, Watts AB, Miller DA, editors. Formulating Poorly Water

Soluble Drugs. New-York: Springer-Verlag; p. 363-442.

Ormes JD, Zhang D, Chen AM, Hou S, Krueger D, Nelson T, Templeton A. 2013.

Design of experiments utilization to map the processing capabilities of a micro-

spray dryer: particle design and throughput optimization in support of drug

discovery. Pharm Dev Technol. 18(1):121-129.

Ousset A, Chavez PF, Meeus J, Robin F, Schubert MA, Somville P, Dodou K. 2018.

Prediction of Phase Behavior of Spray-Dried Amorphous Solid Dispersions:

Assessment of Thermodynamic Models, Standard Screening Methods and a

Novel Atomization Screening Device with Regard to Prediction Accuracy.

Pharmaceutics. 10(1):29.

Parikh T, Gupta SS, Meena AK, Vitez I, Mahajan N, Serajuddin ATM. 2015.

Application of film-casting technique to investigate drug-polymer miscibility in

solid dispersion and hot-melt extrudate. J Pharm Sci. 104(7):2142-2152.

Patel BB, Patel JK, Chakraborty S, Shukla D. 2015. Revealing facts behind spray dried

solid dispersion technology used for solubility enhancement. Saudi Pharm J.

23(4):352-365.

Patel S, Kou X, Hou HH, Huang YB, Strong JC, Zhang GGZ, Sun CC. 2017.

Mechanical Properties and Tableting Behavior of Amorphous Solid Dispersions.

J Pharm Sci. 106(1):217-223.

Paterson AHJ, Brooks GF, Bronlund JE, Foster KD. 2005. Development of stickiness in

amorphous lactose at constant T−Tg levels. Int Dairy J. 15(5):513-519.

Patnaik S. 2016. Novel nanoformulations for enhanced oral bioavailability of some

poorly water soluble non steroidal anti inflammatory drugs. Andhra Pradesh: Sri

Sathya Sai Institute of Higher Learning.

Paudel A, Worku ZA, Meeus J, Guns S, Van den Mooter G. 2013. Manufacturing of

solid dispersions of poorly water soluble drugs by spray drying: formulation and

process considerations. Int J Pharm. 453(1):253-284.

ProCept. 2014. ProCept Information Brochure Spray dryer. Zelzate, Belgium:

http://www.procept.be/spray-dryer-chiller.; [accessed].

Puppolo MM, Hughey JR, Dillon T, Storey D, Jansen-Varnum S. 2017. Biomimetic

Dissolution: A Tool to Predict Amorphous Solid Dispersion Performance.

AAPS PharmSciTech. 18(8):2841-2853.

Rumondor AC, Stanford LA, Taylor LS. 2009. Effects of polymer type and storage

relative humidity on the kinetics of felodipine crystallization from amorphous

solid dispersions. Pharm Res. 26(12):2599-2606.

Schmid K, Arpagaus C, Friess W. 2011. Evaluation of the Nano Spray Dryer B-90 for

pharmaceutical applications. Pharm Dev Technol. 16(4):287-294.

Shamma RN, Basha M. 2013. Soluplus®: A novel polymeric solubilizer for

optimization of Carvedilol solid dispersions: Formulation design and effect of

method of preparation. Powder Technol. 237:406-414.

Singh A, Van den Mooter G. 2016. Spray drying formulation of amorphous solid

dispersions. Adv Drug Deliv Rev. 100:27-50.

Six K, Leuner C, Dressman J, Verreck G, Peeters J, Blaton N, Augustijns P, Kinget R,

Van den Mooter G. 2002. Thermal properties of Hot-stage extrudates of

Itraconazole and Eudragit E100 Phase separation and polymorphism. J Therm

Anal Calorim. 68:591-601.

Six K, Verreck G, Peeters J, Augustijns P, Kinget R, Van den Mooter G. 2001.

Characterization of glassy itraconazole: a comparative study of its molecular

mobility below Tg with that of structural analogues using MTDSC. Int J Pharm.

213:163-173.

Sun DD, Wen H, Taylor LS. 2016. Non-Sink Dissolution Conditions for Predicting

Product Quality and In Vivo Performance of Supersaturating Drug Delivery

Systems. J Pharm Sci. 105(9):2477-2488.

Teja SB, Patil SP, Shete G, Patel S, Bansal AK. 2013. Drug-excipient behavior in

polymeric amorphous solid dispersions. J Excipients and Food Chem. 4(3):70-

93.

Thakral NK, Ray AR, Bar-Shalom D, Eriksson AH, Majumdar DK. 2012. Soluplus--

solubilized citrated camptothecin--a potential drug delivery strategy in colon

cancer. AAPS PharmSciTech. 13(1):59-66.

Tonon RV, Brabet C, Hubinger MD. 2008. Influence of process conditions on the

physicochemical properties of açai (Euterpe oleraceae Mart.) powder produced

by spray drying. J Food Eng. 88(3):411-418.

Ueda K, Higashi K, Yamamoto K, Moribe K. 2014. The effect of HPMCAS functional

groups on drug crystallization from the supersaturated state and dissolution

improvement. Int J Pharm. 464(1-2):205-213.

Van Bockstal PJ, Mortier S, Corver J, Nopens I, Gernaey KV, De Beer T. 2018. Global

Sensitivity Analysis as Good Modelling Practices tool for the identification of

the most influential process parameters of the primary drying step during freeze-

drying. Eur J Pharm Biopharm. 123:108-116.

Van Daele T, Gernaey KV, Ringborg RH, Börner T, Heintz S, Van Hauwermeiren D,

Grey C, Krühne U, Adlercreutz P, Nopens I. 2017. Application of iterative

robust model-based optimal experimental design for the calibration of

biocatalytic models. Biotechnol Prog. 33(5):1278-1293.

Vasconcelos T, Sarmento B, Costa P. 2007. Solid dispersions as strategy to improve

oral bioavailability of poor water soluble drugs. Drug Discov Today. 12(23-

24):1068-1075.

Vicente J, Pinto J, Menezes J, Gaspar F. 2013. Fundamental analysis of particle

formation in spray drying. Powder Technol. 247:1-7.

Wang B, Xu DL, Chu KW, Yu AB. 2006. Numerical study of gas–solid flow in a

cyclone separator. Appl Math Model. 30(11):1326-1342.

Appendices

Appendix 1: Response surface DoE and yield results obtained for the 100 mg

productions of ASD of Itraconazole at DL 40% (w/w).

Run Drying airflow

(m³/min)

Inlet temperature

(°C)

Pump speed

(g/min)

Atomizing airflow

(bars) Polymers

Yield

(%)

1 0.33 60 1 1 HPMCAS-LF 51.6

2 0.45 60 4.5 1 Soluplus 70.7

3 0.20 60 1 0.5 Soluplus 35.0

4 0.33 60 8 0.5 Soluplus 48.4

5 0.20 60 8 1.5 Soluplus 48.4

6 0.20 60 8 0.5 HPMCAS-LF 28.8

7 0.33 60 1 1.5 Soluplus 61.3

8 0.20 60 4.5 1.5 HPMCAS-LF 37.7

9 0.45 60 1 0.5 HPMCAS-LF 44.6

10 0.45 60 8 1.5 HPMCAS-LF 54.1

11 0.33 90 4.5 1 Soluplus 46.4

12 0.33 90 4.5 1 Soluplus 36.4

13 0.45 90 4.5 0.5 Soluplus 43.8

14 0.33 90 4.5 1 HPMCAS-LF 47.6

15 0.33 90 4.5 1 HPMCAS-LF 33.6

16 0.45 90 1 1 HPMCAS-LF 47.5

17 0.45 90 8 0.5 HPMCAS-LF 19.2

18 0.20 90 1 0.5 HPMCAS-LF 43.2

19 0.33 90 8 1.5 HPMCAS-LF 36.2

20 0.45 90 1 1.5 Soluplus 59.6

21 0.33 90 4.5 1 Soluplus 42.0

22 0.20 120 8 1 HPMCAS-LF 32.0

23 0.33 120 1 0.5 HPMCAS-LF 27.3

24 0.45 120 1 0.5 Soluplus 49.8

25 0.20 120 4.5 1.5 Soluplus 49.9

26 0.2 120 1 1.5 HPMCAS-LF 46.6

27 0.45 120 8 1 Soluplus 46.2

28 0.2 120 1 1 Soluplus 43.3

29 0.45 120 4.5 1.5 HPMCAS-LF 34.9

30 0.33 120 4.5 0.5 HPMCAS-LF 31.5

31 0.2 120 8 0.5 Soluplus 23.1

32 0.2 120 8 1 HPMCAS-LF 27.9

33 0.33 120 8 1.5 Soluplus 48.5

Tables

Table 1: Physicochemical properties of selected polymers.

Polymer Mw (g/mol) Dissolution pH Tg (°C)

HPMCP HP50 78000 > 5.0 140

HPMCAS-LF 18167 > 5.5 122

PVPVA 57500 - 112

PVPK30 50000 - 162

Soluplus 115000 - 70

Eudragit L100 125000 > 6.0 192

Eudragit L100-55 320000 > 5.5 122

Table 2: Level values of process and formulation parameters included in the DoE.

Factors Unit Low level Center level High level

Drying airflow m³/min 0.2 0.33 0.45

Temperature inlet °C 60 90 120

Atomizing airflow bars 0.5 1 1.5

Pump speed g/min 1 4.5 8

Polymer type - HPMCAS-LF - Soluplus

Table 3: List of process parameters, formulation attributes and process configuration

kept constant in the DoE.

Factors Unit Value/attribute/configuration

Nozzle orifice diameter mm 1.2

Cooling airflow L/min 100

Cyclone size - Medium

Solid content mg/mL 50

DL % (w/w) 40

Drug - Itraconazole

Solvent - DCM/EtOH 2:1 (v/v)

Table 4: Optimized values of process parameters identified by the model and adapted

for the small cyclone use.

Factors Unit Medium cyclone Small cyclone

Drying airflow m³/min 0.45 0.3

Inlet temperature °C 60 60

Atomizing airflow bars 1.5 1.5

Pump speed g/min 1 1

Table 5: Mean values of yield (%) and standard deviation (SD) (%) for solid dispersions

of Itraconazole and Naproxen produced with the medium cyclone at a fixed DL of 40%

(w/w).

Drug DL Cyclone Polymers HPMCP

HP50

HPMCAS

-LF PVPVA PVPK30 Soluplus

Eudragit

L100

Eudragit

L100-55

Itraconazole 40% Medium

Yield (%) 54.4 54.0 53.4 49.0 56.6 59.6 53.5

SD (%) 3.5 8.4 6.9 5.5 3.6 8.2 7.0

Naproxen 40% Medium

Yield (%) 53.8 56.3 44.4 47.6 49.8 55.6 49.5

SD (%) 2.7 2.7 4.1 4.4 2.2 2.6 3.9

Table 6: Miscibility characterization of ASDs produced by spray-drying at 2.5 g,

100 mg and 20 mg batch size and prepared by solvent casting and quench cooling

for Itraconazole (a) and Naproxen (b) ASDs.

a) Itraconazole ASDs DL

(w/w) HPMCP

HP50

HPMCAS

-LF PVPVA PVPK30 Soluplus

Eudragit

L100

Eudragit

L100-55

Spray-drying -

2.5 g

20%

GS GS

GS GS

DM/DR GS GS 40% PS PS

Small scale

Spray-drying –

100 mg

20% GS GS

GS GS DM/DR GS GS

40% PS PS

Small scale

Spray-drying –

20 mg

20% GS GS

GS GS PS GS GS

40% PS PS DM/DR

Solvent casting 20% GS GS

DM/DR DM/DR DM/DR GS GS 40% PS PS

Quench cooling 20%

GS GS GS GS GS GS GS 40%

b) Naproxen ASDs DL

(w/w) HPMCP

HP50

HPMCAS

-LF PVPVA PVPK30 Soluplus

Eudragit

L100

Eudragit

L100-55

Spray-drying -

2.5 g

20%

DM/DR DM/DR GS GS

GS GS

DM/DR 40% DM/DR DM/DR

Small scale

Spray-drying –

100 mg

20% DM/DR DM/DR GS GS

GS GS DM/DR

40% DM/DR DM/DR

Small scale

Spray-drying –

20 mg

20%

DM/DR DM/DR GS GS

GS GS

DM/DR 40% DM/DR DM/DR

Solvent casting 20%

DM/DR GS PS GS

GS DM/DR DM/DR 40% DM/DR DM/DR DM/DR

Quench cooling 20%

GS GS GS GS GS GS GS

40% DM/DR DM/DR

Miscibility characterization examined by mDSC;

GS: glass solution, PS: Phase separation, DM/DR: drug melting/drug recrystallization

Table 7: Summary of stability results of 40:60 (w/w) solid dispersions of

Itraconazole and Naproxen upon storage at 25°C/60% RH and 40°C/75% RH.

Drug Carriers DL

(w/w)

25°C/60% RH 40°C/75%RH

2w 5w 2w 5w

Itraconazole HPMCP HP50 40:60 N N N N

HPMCAS-LF 40:60 N N N N

PVPVA 40:60 N N N Y

PVPK30 40:60 N N N Y

Soluplus 40:60 N N Y Y

Eudragit L100 40:60 N N N N

Eudragit L100-55 40:60 N N N N

Naproxen PVPVA 40:60 N N N N

PVPK30 40:60 N N N N

Residual crystallinity examined by XRPD; N: No crystalline signal, Y: crystalline signal

Figures

Figure captions

Figure 1: Fit model analysis for yield prediction.

Figure 2: Maximized yield response for 100 mg production of spray-dried ASDs of

Itraconazole mixed with HPMCAS-LF (A) and Soluplus (B) at 40% (w/w) DL.

Confidence interval at 95% is represented in dash blue lines.

Figure 3: Distribution map of mean yield values as a function of pump speed with

atomizing airflow, drying airflow and inlet temperature respectively, for overlapped 100

mg ASD productions of Itraconazole/HPMCAS and Itraconazole/Soluplus 40:60 (w/w)

with optimized process conditions.

Figure 4: Yield distribution based on simulation of 5000 productions computed for the

40:60 (w/w) ASD of Itraconazole mixed with HPMCAS-LF (A) and Soluplus (B);

under identified process conditions.

Figure 5: SEM microphotographs of 40:60 (w/w) Itraconazole SDSDs with:

HPMCP HP50 (a), HPMCAS-LF (b), PVPVA (c), PVPK30 (d), Soluplus (e) and

Eudragit L100 (f) collected from the small cyclone.

Figure 6: Average yield values (n=3) obtained for the 100 mg ASD productions of

20:80 and 40:60 (w/w) Itraconazole (a) and Naproxen (b) using the small cyclone. The

minimum yield is represented by the solid red line at 50%.

Figure 7: Cyclone efficiency: yield improvement (%) for the 100 mg ASD productions

of 40:60 (w/w) Itraconazole and Naproxen performed under optimized process

parameters using small/medium cyclone.

Figure 8: Average yield values (n=3) obtained for the 20 mg ASD productions of 20:80

and 40:60 (w/w) Itraconazole (a) and Naproxen (b) using the small cyclone. The

minimum yield is represented by the solid red line at 10%.

Figure 9: Reverse heat flow signals of Itraconazole/PVPVA 40:60 (w/w) produced by

spray-drying for 2.5 g batch production (a), small-scale spray-drying approach for 20

mg (b) and 100 mg (c) batches production, quench cooling (d) and solvent casting (e).

The arrows indicated the presence of the Tg, the mesophases of glassy Itraconazole and

the drug melting endotherm.

Figure 10: XRPD patterns of 40:60 SDSDs of Itraconazole and Naproxen just after

processing and after 5 weeks under stress conditions at 40°C/75% RH, respectively

with HPMCP HP50 (a), HPMCAS-LF (b), PVPVA (c), PVPK30 (d), Soluplus (e),

Eudragit L100 (f) and Eudragit L100-55 (g).

Figure 11: Dissolutions profiles (a) and solubility improvement (%) (b) of screened

40:60 (w/w) Itraconazole SDSDs compared to crystalline drug after 60 and 150

minutes in 50 mM phosphate buffer (pH 6.5) containing 2% Vitamine E TPGS at a

drug concentration of 1 mg/mL.

Figure 12: Solubility improvement (%) of screened 40:60 (w/w) Naproxen SDSDs

compared to crystalline drug after 60 and 150 minutes in 50 mM phosphate buffer

(pH 6.5) containing 0.5% SDS and 2% HPMC at a drug concentration of 5mg/mL.

Figure 13: Dissolutions profiles (a) and solubility improvement (%) (b) of screened

40:60 (w/w) Naproxen SDSDs compared to crystalline drug after 60 and 150

minutes in simulated gastric fluid (pH 1.2) containing 0.5% SDS and 2% HPMC at

a drug concentration of 1 mg/mL.